GNAS mutation inhibits growth and induces phosphodiesterase 4D expression in colorectal cancer cell lines

Approximately 5% of colorectal cancers (CRCs) have a gain‐of‐function mutation in the GNAS gene, which leads to the activation of cAMP‐dependent signaling pathways and associates with poor prognosis. We investigated the effect of an activating GNAS mutation in CRC cell lines on gene expression and cell proliferation in vitro, and tumor growth in vivo. GNAS‐mutated (GNASmt) HCT116 cells showed stimulated synthesis of cAMP as compared to parental (Par) cells. The most upregulated gene in the GNASmt cells was cAMP‐hydrolyzing phosphodiesterase 4D (PDE4D) as detected by RNA sequencing. To further validate our finding, we analyzed PDE4D expression in a set of human CRC tumors (n = 35) and demonstrated overexpression in GNAS mutant CRC tumors as compared to GNAS wild‐type tumors. The GNASmt HCT116 cells proliferated more slowly than the Par cells. PDE4 inhibitor Ro 20‐1724 and PDE4D subtype selective inhibitor GEBR‐7b further suppressed the proliferation of GNASmt cells without an effect on Par cells. The growth inhibitory effect of these inhibitors was also seen in the intrinsically GNAS‐mutated SK‐CO‐1 CRC cell line having high levels of cAMP synthesis and PDE4D expression. In vivo, GNASmt HCT116 cells formed smaller tumors than the Par cells in nude mice. In conclusion, our findings demonstrate that GNAS mutation results in the growth suppression of CRC cells. Moreover, the GNAS mutation‐induced overexpression of PDE4D provides a potential avenue to impede the proliferation of CRC cells through the use of PDE4 inhibitors.

cultured CRC cells, they showed that GNAS mutations induced cells to make more cAMP and cAMP-hydrolyzing enzymes.Inhibiting these enzymes suppressed CRC cell proliferation, suggesting that these inhibitors could potentially be used to slow the growth of CRC tumors with GNAS mutations.

| INTRODUCTION
The intracellular level of cyclic adenosine monophosphate (cAMP) is tightly controlled by the balance of production by adenylyl cyclase (AC) and degradation by cyclic nucleotide phosphodiesterase (PDE). 1 The upstream signal that stimulates the synthesis of cAMP is largely provided by the stimulatory α subunit (Gα s ) of heterotrimeric guanine nucleotide-binding protein (G protein).Activated Gα s transduces signals from G protein-coupled receptors (GPCRs) to AC, which becomes activated and produces cAMP.Gα s subunit is encoded by the GNAS (guanine nucleotide binding protein, alpha stimulating) gene, and activating mutations of this gene lead to reduced intrinsic GTPase activity of the Gα s , rendering it and cAMP-dependent signaling pathways to become activated. 1,2In addition to the major effector protein kinase A (PKA), cAMP is capable of activating other effectors, including exchange protein directly activated by cAMP (EPAC), cyclic nucleotide-gated ion channels and Popeye domain containing proteins. 1,3ere is a striking enrichment of Gα s -pathway activation in cancers of the gastrointestinal tract, and over half of the colorectal cancers (CRC) have aberrantly turned on this pathway by activation of GPCRs or by gene mutations and rearrangements. 1,2The occurrence of GNAS mutations in CRC is 4.8% according to a recent metaanalysis, most of them being either R201C or R201H, which associate with poor prognosis and resistance to oncological treatments. 4,5However, the role of GNAS mutations in gastrointestinal carcinogenesis is somewhat contradictory, since they are predominantly found in precursor lesions and are thought to be important only in the initiation phase of carcinogenesis. 1 In addition to activation of Gα s , it is intriguing that PDE dysregulation also occurs in CRC. 1,6Es hydrolyze the phosphodiester bond of cAMP and/or cGMP, yielding non-cyclic nucleotides unable to transmit downstream signaling. 7together 21 human PDE genes are known, and PDE enzymes are subdivided into 11 families (PDE1-11) based on sequence homology, enzymatic properties and sensitivity to inhibitors.Each family contains several transcript variants with unique tissue and subcellular distribution, regulation of expression and post-translational modification. 8Expression of two cAMP-specific PDE subtypes, PDE4B and PDE4D, has been shown to be elevated in CRC and modulate the growth and survival of CRC cells. 6While there exist no pharmacological inhibitors for Gα s , PDE inhibitors have been investigated as potential therapeutic agents in various diseases.Some of these inhibitors have already been approved for the treatment of cardiovascular diseases (PDE3 inhibitors), pulmonary hypertension and erectile dysfunction (PDE5 inhibitors) and inflammatory conditions (PDE4 inhibitors). 9There is an increasing interest in PDE inhibitors as a therapeutic option for malignant diseases. 1,6,9[12][13][14] In this study, we investigated the effect of an activating GNAS mutation in CRC cell lines on gene expression, cell proliferation in vitro and tumor growth in vivo.We show that GNAS mutationinduced effects in CRC cells are counteracted by overexpression of PDE4D, which can be reversed by using PDE4 inhibitors.

| siRNA transfection
To knock down GNAS expression, HCT116 and SK-CO-1 cells were transfected with siRNAs utilizing Lipofectamine™ RNAiMAX transfection reagent (Thermo Fisher Scientific) according to the manufacturer's protocol, using 10 nM siRNA concentration, OPTI-MEM I reduced serum medium and 48 or 72 h transfection time.The siRNAs used for GNAS knockdown were s526399 (siRNA #2) and s5891 (siRNA #4), while Silencer Select GAPDH Positive control siRNA and Silencer Select Negative control #2 siRNA were used as controls.All experiments were repeated at least three times with two or three biological replicates.

| cAMP assay
Basal cAMP synthesis/accumulation rates of the cell lines were investigated using cAMP-Glo Max Assay (Promega, Madison, WI) according to the kit instructors.For the assay, 8000 cells were seeded into Corning BioCoat poly-D-lysine-treated 96-well plates the previous day, and at the start of the assay the cells were incubated for 2 to 3 h in Complete Induction Buffer (containing 500 μM pan-PDE inhibitor IBMX and 100 μM PDE4 inhibitor Ro 20-1724 to enable cAMP accumulation).The experiments were repeated at least three times with four to six replicates.

| RNA-Seq analysis
Total RNA was extracted from cell pellets using NucleoSpin RNA kit  S1.

| Reverse transcription-PCR
PDE4D transcript variants expressed in the studied cell lines were analyzed using reverse transcription (RT)-PCR.For PCR, 1/10 vol of cDNA synthesized as above, dNTP mix, DreamTaq DNA polymerase (Thermo Fischer Scientific) and isoform-specific primers (Eurofins Genomics, Ebersberg, Germany) were mixed and amplified using standard PCR protocol.Details of the primers and PCR conditions are presented in Table S2.Primer sequences for PDE4D7 were from Wang et al 16 and for other isoforms from Lin et al. 11 The PCR products were finally resolved in 2% agarose gel, using Midori Green Advance DNA Stain (Nippon Genetics Europe GmbH, Düren, Germany) for visualization.A549 total RNA and human brain total RNA (Takara Bio Europe, Saint-Germain-en-Laye, France) were used as positive controls, and actin beta (ACTB) primers were used as a loading control.
Relative cell numbers were calculated from fluorescence measured at 0, 48 and 72 h after the addition of the inhibitors by using the CellTiter-Blue Cell Viability Assay (Promega).The redox dye resazurincontaining reagent was incubated for 1 h before measurement of fluorescence (Ex 544 nm, Em 590 nm) by FLUOstar Omega plate reader (BMG Labtech, Ortenberg, Germany).All the experiments were repeated at least three times with four to six replicates.

| Thymidine incorporation assay
As a second method to inspect cell proliferation, a 3 H-thymidine incorporation assay was used.Briefly, 50,000 cells were seeded into six-well plates and allowed to grow overnight.The next day, the medium was changed to fresh one containing DMSO (0.05%; Ctrl), GEBR-7b (10 μg/mL), Ro 20-1724 (10 μM) or FSK (10 μM) and the cells were allowed to proliferate for 24, 48 or 72 h.Four hours before each timepoint, 0.4 μCi/mL 3 H-thymidine (PerkinElmer, Waltham, MA) was added.At each time point, the cells were washed three times with PBS and incubated for 10 min with 5% trichloroacetic acid and then 10 min with 0.1 M NaOH.Finally, the samples were transferred into scintillation tubes, high sample load scintillation cocktail Optiphase Hisafe 3 (PerkinElmer) was added, and radioactivity was measured using Wallac 1414 liquid scintillation counter (PerkinElmer).All experiments were repeated at least three times with three biological replicates.

| Apoptosis assay
For analysis of apoptosis, Caspase-Glo 3/7 Assay (Promega) was performed according to the manufacturer's instructions.For the assay, 7500 HCT116 or 10000 SK-CO-1 cells were seeded into white clear bottom 96-well plates (3610; Corning), and PDE4 inhibitors, FSK or DMSO Ctrl were added the next day.The concentrations used were the same as above.Staurosporine (1 or 5 μM) was used as a positive control.After the addition of the inhibitors, the cells were incubated for 24 h.The Caspase-Glo 3/7 Reagent was added and incubated for 1 h before measuring the luminescence with FLUOstar Omega plate reader (BMG Labtech).All the experiments were repeated at least three times with three to four replicates.

| In vivo xenograft assays
For animal experiments, 4-weeks old female NMRI-Foxn1 nude mice were purchased from Janvier laboratories (www.janvier-labs.com) and maintained in ventilated caging rack system under a pathogen free environment in the University of Helsinki Biomedicum Animal Facility, in accordance with the ethical guidelines of the provincial government of Finland.For in vivo tumor growth assay, 1 Â 10 6 HCT116 Par and GNASmt cells in 1:1 diluted Matrigel were implanted subcutaneously into the flanks of the immunodeficient mice (one flank per animal, eight mice per study group).Starting from day seven, the tumor sizes were measured twice per week using a digital caliper 0.150 mm (Cocraft stainless hardened) and all animals were sacrificed when the largest tumor reached ethical limits.The weight of the mice was measured twice a week.Tumor data were normalized to day seven, which was set as 100%.
2.13 | Analysis of PDE4D expression in CRC tumors stratified by GNAS mutation status PDE4D mRNA expression was analyzed from an existing whole genome sequencing (WGS) and RNA-Seq data of CRC material (n = 34) described in Cajuso et al. 17 Further, one extra sample analyzed in similar way was included.WGS mutation statuses were compiled with Base-Player. 18Salmon (version 0.12.0;quant mode with --validateMappings) was used to map raw sequences onto the human transcriptome (Ensembl release 79).Gene-level quantification and differential gene expression (DE) analysis were done with DESeq2 (v1.18.1) using default options.Variance stabilizing transformation was used to estimate logarithmic-scale values for gene expression visualization.For DE, we report fold-change (FC) and P-value (DESeq2 Wald test).

| Immunohistochemistry
The mouse xenograft tumors were cut into two halves, fixed for 24 h with formalin, transferred into 70% ethanol for at least 1 h and embedded in paraffin using Tissue-Tek TEC Embedding System (Sakura Finetek, Tokyo, Japan).HE staining was performed with a routine protocol.For analysis of cell proliferation (Ki67) and apoptosis (cleaved caspase 3), the sections were deparaffinized and rehydrated, subjected to heatinduced epitope retrieval (HIER) in 10 mM Tris, 1 mM EDTA (pH 9.0) and incubated 1 h at RT with the primary antibodies (rabbit anti-Ki67, ab92742, Abcam or rabbit anti-Cleaved Caspase 3, #9661, Cell Signaling Technology, Danvers, MA) using dilutions 1:500 and 1:200, respectively.
For detection, BrightVision goat anti-rabbit HRP (DPVR-110HRP, Immunologic, Duiven, the Netherlands) was incubated for 30 min at RT and BrightDAB (BS04-110; Immunologic) for 8 min at RT.The percentages of immunopositive cells were determined from six high-power fields (Â400) of the vital tumor areas, counting positive nuclei for Ki67 staining and positive cytoplasm for cleaved caspase 3 staining.

| Statistical analyses
All the graphs show results as mean ± SEM.Statistical significance (P < .05)for the differential gene expression by RNA-Seq was evaluated by Benjamini-Hochberg adjusted P-values using R.The difference in the in vitro assays was evaluated by the two-tailed Mann-Whitney U test using Prism 9 (GraphPad Software, San Diego, CA).
For the in vivo experiment, Sidak's multiple comparison tests were analyzed with 2-way ANOVA using Prism 9 (GraphPad Software).

| GNAS mutation stimulates cAMP synthesis in HCT116 CRC cells
GNASmt HCT116 cells showed increased cAMP synthesis as compared to Par cells (Figure 1A).Knock down of GNAS expression by GNAS siRNAs blocked the GNAS mutation-induced synthesis of cAMP, without any effect on the Par cells (Figure 1B).The effect of GNAS siRNAs on GNAS mRNA expression is shown in Figure S1A.

| GNAS-mutated HCT116 cells show striking upregulation of PDE4D expression
GNASmt HCT116 cells showed altogether 159 significantly upregulated genes and 130 downregulated genes when compared to the Par cells as analyzed by RNA-Seq and using FC 2.0 as a cutoff (Table S3).The most highly induced gene was PDE4D (FC > 700, P < .00001),which was clearly the most upregulated of any cAMP-cleaving PDEs (Table 1).
Although PDE4B expression was relatively high in the Par cells, it was only modestly (1.5-fold) induced in the GNASmt cells (Table 1).The prominent upregulation of PDE4D, as well as the moderate increase in PDE4B, was verified with qPCR (Figure 2A,B).The strong induction of PDE4D expression in the GNASmt cells was effectively downregulated by GNAS siRNAs (Figure 2C), while they showed no effect or only a modest effect on PDE4B expression (Figure 2D).Among the transcript variants of PDE4D, induction seemed to specifically target the short and supershort variants PDE4D1, PDE4D2 and PDE4D6 (Figure S2A).This was supported by protein-level data obtained by using Western blotting (Figure S2B).Related to the potential downstream signaling mediating  the effect of cAMP, PKA and cAMP response element-binding protein (CREB) inhibitors effectively downregulated PDE4D transcript expression (41% and 53% inhibition, respectively), whereas EPAC inhibitor had a minor effect (22% inhibition) (Figure S3).Based on RNA-Seq data, PKA subunits also showed higher expression levels in GNASmt cells than EPACs, PRKACs and PRKARs giving thousands of reads (PRKACB mean value being 11,000 and PRKAR1A mean value 15,000), whereas the values for both RAPGEF3 and RAPGEF4 (coding for EPAC1 and EPAC2, respectively) were below 100 (Table S3 and data not shown).

| PDE4 inhibition suppresses the proliferation of GNASmt HCT116 cells
GNASmt cells proliferated more slowly than the Par cells (28% inhibition as compared to the Par cells at 72 h, P < .001) as analyzed by metabolic activity-dependent cell viability assay (Figure 3A).PDE4 selective inhibitor Ro 20-1724 or PDE4D subtype selective inhibitor GEBR-7b further inhibited the proliferation of GNASmt cells (42% and 45% inhibition, respectively, as compared to the Par cells at 72 h, P-values <0.001), without an effect on Par cells (Figure 3A).FSK showed a similar effect on the proliferation of GNASmt cells (46% inhibition as compared to the Par cells at 72 h, P < .001; Figure 3A).The proliferation impairment was further verified with thymidine incorporation assay.In this assay, GNASmt cells showed a 22% inhibition as compared to the Par cells, and PDE4 inhibitors showed a 40% to 41% suppression as compared to the Par cells in a 72 h experiment (Figure 3B).FSK inhibited thymidine incorporation in both Par and GNASmt cells by 19% and 60%, respectively, as compared to the Par Ctrl cells at the 72 h time point (Figure 3B).Our results suggest that cell growth is here the key target of GNAS mutation, since it did not induce apoptosis when compared to the Par cells in caspase 3/7 activity assay (Figure S4A).Furthermore, Ro 20-1724 did not induce apoptosis in Par or GNASmt cells, and GEBR-7b and FSK induced only a slight increase

| GNASmt HCT116 cells form smaller tumors than the Par cells in mouse xenografts
In line with the in vitro findings, GNASmt HCT116 cell-derived tumors were smaller than those of Par cells, when grown subcutaneously in nude mice (Figure 4A).Additionally, the percentage of Ki67 immunopositive cells was significantly lower in the GNASmt tumors as compared to the Par tumors (93.6% ± 0.5% vs 95.6% ± 0.5%, P = .02).However, cleaved caspase 3 immunostaining revealed slightly more positivity in the Par tumors than in the GNASmt tumors (3.6% ± 0.3% vs 2.5% ± 0.2%, P = .01).

| GNAS-mutated human CRC tumors show high expression of PDE4D
To further validate the relationship between GNAS mutation and PDE4D expression, we analyzed PDE4D mRNA expression in GNAS  wild-type and mutant human CRC tumors.We identified four cases with GNAS codon mutation (three R201H and one R201C) in a series of 35 CRCs, and all GNAS mutant tumors (n = 4) showed a high expression of PDE4D (Figure 4B).The median value of PDE4D expression for GNAS wild-type tumors (n = 31) was 8.33 (interquartile range, IQR 8.01-8.65),and PDE4D expression of all GNAS mutant tumors (n = 4) were in the fourth quartile of the wild-type tumors.Overall, PDE4D mRNA expression was upregulated in GNAS mutant tumors as compared to wild-type ones (FC = 1.8, P = .014).Further, three of the four GNAS mutant tumors showed moderate to strong immunostaining for PDE4D in the neoplastic cells (Figure S5).In addition to PDE4D mRNA, PDE4B mRNA expression showed higher values in GNAS mutant tumors, but the difference between them and GNAS wild-type tumors was not statistically significant (FC = 2.3, P = .05)(Figure S6).All four GNAS mutant tumors also contained a RAS pathway mutation (BRAF V600E in two, KRAS G13D in one and KRAS A146V in one), and two of them contained APC frameshift mutation.
The four GNAS wild-type tumors showing the highest PDE4D expression did not contain PRKACA or PRKAR1A mutations, or GNAS or PDE4D gene fusions.Three of them contained KRAS codon 12 mutation (G12D, G12S and G12C) and one contained NRAS codon 12 mutation (G12V).None of the analyzed 35 CRC tumors contained mutations in the coding regions of PDE4D or PDE4B genes.Interestingly, both PDE4D and PDE4B expression showed a positive correlation with cyclooxygenase 2 (COX-2) gene (PTGS2, ENSG00000073756) expression in GNAS wild-type tumors (Pearson correlation r = .52,P = .003for PDE4D and r = .50,P = .004for PDE4B, n = 31 in both).GPCR activation due to COX-2 produced prostaglandin E 2 (PGE 2 ) might thus explain the high PDE4D/B expression seen in some of the GNAS wild-type tumors.

| GNAS mutant SK-CO-1 CRC cell line is sensitive to PDE4 inhibition
To verify our findings in a CRC cell line intrinsically having an activating GNAS mutation, we utilized GNAS R201C mutated SK-CO-1 cell line.SK-CO-1 cells showed a high cAMP synthesis rate and a high PDE4D expression level (Figure 5A,B).PDE4B expression level, in turn, was low in SK-CO-1 cells (Figure 5C).Similarly to GNASmt HCT116 cells, PDE4D expression in SK-CO-1 cells was downregulated by GNAS siRNAs (Figure 5D).In contrast to GNASmt HCT116 cells, in this cell line also PDE4B expression was downregulated by GNAS (Figure 5E).The effect of GNAS siRNAs on GNAS mRNA expression is shown in Figure S1B.Importantly, PDE4 inhibition or FSK inhibited the proliferation of SK-CO-1 cells (Figure 5F).PDE4 inhibitors did not induce apoptosis in these cells, but FSK induced a 2-fold increase in caspase 3/7 activity (Figure S4B).As a positive control, 1 μM staurosporine induced over 6-fold caspase activation (data not shown).SK-CO-1 cell line did not grow in the nude mouse model (<300 mm 3 tumors in 42 days after inoculating 4 Â 10 6 cells in Matrigel; data not shown).

| DISCUSSION
We investigated the molecular and functional effects of mutated GNAS oncogene in CRC cell lines.Synthesis of cAMP was highly stimulated in GNASmt HCT116 cells as compared to the GNAS wild-type Par cells, which is in line with a previous publication using HT-29 CRC cells. 19The topmost gene expression change induced by the GNAS mutation was a striking several hundred-fold induction of the cAMP-cleaving phosphodiesterase PDE4D.In contrast to PDE4D, which was expressed at a very low level in the Par HCT116 cells, PDE4B was highly expressed in the Par cells, which is consistent with earlier reports. 20,21However, we show that expression of PDE4B was only slightly induced in the GNAS-mutated cells.Further, in the inherently GNAS-mutated SK-CO-1 cell line, showing even higher cAMP synthesis than the GNASmt HCT116 cells, the expression of PDE4D was high and that of PDE4B low.We also show that knockdown of GNAS downregulated expression of PDE4D in GNASmt HCT116 and SK-CO-1 cells, which indicates that activation of GNAS is the key regulator of PDE4D expression in these cells.The likely mechanism for PDE4D overexpression in GNAS mutant cells is the increase in cAMP production that leads to activation of PKA and CREB-mediated stimulation of cAMP response element (CRE) in the PDE4D gene promoter. 16,22cAMP-induced PDE4D expression thus constitutes a compensatory mechanism to counteract the prolonged stimulation of cAMP synthesis caused by activating GNAS mutation.To support this, we obtained effective downregulation of PDE4D expression by PKA and CREB inhibitors, whereas an EPAC inhibitor was less effective.
Finally, we demonstrated a high expression of PDE4D transcript and protein in GNAS-mutated human CRC tumor specimens as compared to GNAS wild-type ones, which indicates an important association between mutational activation of GNAS and PDE4D expression in human tumors as well.This is also supported by results published by Arang and Gutkind (n = 10,437), 2 who showed a higher than average prevalence of GNAS mutations (endometrial 7.4%, gastric 5.7%, pancreatic 5.0%, colon 4.5% and lung carcinoma 3.7%, as well as cutaneous melanoma 6.1%, vs pan-cancer average 2.3%) in cancer types reported to overexpress PDE4D. 11,12,14ASmt HCT116 cells were less proliferative than the Par cells in vitro, and PDE4 inhibitors further suppressed the proliferation of GNASmt cells without an effect on the Par cells.The growth inhibitory effect of PDE4 inhibitors was also evident in the inherently GNAS mutant SK-CO-1 CRC cell line.Effect of PDE4 inhibitors on cell growth were demonstrated by both cell viability assay and thymidine incorporation assay.However, apoptosis was not induced by the PDE4 inhibitor Ro 20-1724, and negligible effect of another PDE4 inhibitor resveratrol on apoptosis has been demonstrated earlier in 2D cultures of HCT116 cells.23 These data indicate that upregulation of PDE4D provides a growth advantage for GNAS-mutated CRC cells by modulating cAMP levels, and its pharmacologic inhibition leads to suppressed growth.As demonstrated by Li et al, 24 a potential mechanism for cAMP-induced growth suppression of HCT116 cells may depend on PKA-induced phosphorylation of BRAF and CRAF and a concomitant inhibition of ERK activation.In line with our results, CRISPR/Cas9-mediated GNAS R201H silencing has been shown to increase in vitro growth of a GNAS-and KRAS-mutated human pancreatic adenocarcinoma cell line.25 Further, pharmacological inhibition of PDE4D or its gene silencing have been shown to suppress in vitro growth in multiple cancer cell lines, including breast, endometrial, gastric, hepatocellular, lung, ovarian and prostate carcinoma, as well as melanoma.10,11,13 More et al 26  growth rate of prostate cancer cell lines.10 One explanation for the discrepant results between peritoneal and subcutaneous in vivo models may depend on a different microenvironment.
GNAS and KRAS oncogenes are often co-mutated in tumors, 27,28 and both HCT116 and SK-CO-1 CRC cell lines are KRAS-mutated. 29eviously, PDE4 inhibitors have been shown to require combination with FSK to decrease proliferation of KRAS-mutated and GNAS wildtype DLD1 and HCT116 CRC cells. 21,30Similarly, PDE4 inhibition was not effective alone, but again required combination with FSK to inhibit proliferation of GNAS mutant and KRAS/BRAFV600 wild-type KM12C cells. 31In our experiments, Par HCT116 did not respond to PDE4 inhibition alone, but in both the GNASmt HCT116 cells and the inherently GNAS mutant SK-CO-1 cells PDE4 inhibitors suppressed the cell growth as a single agent.To this end, it is interesting to note that half of the human GNAS-mutated CRC tumors in our series had a KRAS mutation.As the other half of these tumors contained BRAF V600E mutations, it is plausible to conclude that combination of GNAS and RAS or BRAF V600E mutations is necessary to increase cAMP to such levels that leads to overexpression of PDE4D.It is also interesting to note that CRC cell lines show enrichment of microsatellite instability (MSI) as compared to CRC tumors, and also HCT116 cell line is MSI. 29However, the SK-CO-1 cell line is microsatellite stable, as were all GNAS-mutated human CRC tumor specimens in our series, indicating that the relationship between GNAS activation and PDE4D overexpression takes place in both microsatellite stable and unstable conditions.
The PDE4D gene gives rise to the highest number of transcript variants of the PDE4 genes, and these variants differ markedly in their N-terminal regions. 8PDE4D1-9 protein isoforms are enzymatically active, whereas PDE4DN1-3 are inactive due to the missing parts of catalytic domain (so-called dead-short variants).Of the enzymatically active isoforms, PDE4D1 is short, PDE4D2 and D6 are supershort, and the rest are long isoforms.Long isoforms can form dimers and they can be activated by PKA through phosphorylation, 32 which is an important negative regulation loop for cAMP signaling.ERK-mediated phosphorylation in turn inhibits the activity of most PDE4D isoforms, except the short isoform D1, which is conversely activated. 33Interestingly, GNAS mutation induced the expression of short and supershort variants PDE4D1, PDE4D2 and PDE4D6 in the HCT116 cells.These variants are known to be transcriptionally upregulated by PKA activated CREB binding to the CRE in their promoter. 34In several cell types, including Jurkat T-cells, NIH3T3 cells, and activated vascular smooth muscle cells, cAMP elevation has been found to specifically induce expression of the short and supershort variants PDE4D1 and PDE4D2. 35,36In addition, epigenetic regulation may also play a role, since histone acetylation of the intronic promoter regulating PDE4D1 and PDE4D2 expression was shown to allow selective cAMP-mediated induction of these transcript variants. 36In our model, at least in HCT116 cells, it seems evident that of the PDE4D isoforms the short and supershort variants play the major role in regulation of cAMP levels and subsequent growth control of these cancer cells.
In conclusion, we show that activating GNAS mutation induces cAMP synthesis and results in growth suppression of CRC cells.Clearly, other cellular mechanisms beyond proliferation are responsible for GNAS gene's oncogenic potential.Importantly, our study underlines the importance of compensatory PDE4 upregulation to provide a growth advantage to GNAS-mutated CRC cells, and the pharmacological use of PDE4 inhibitors could potentially impede the growth of GNAS-mutated tumors.
findings of this study are available from the corresponding author upon request.

ETHICS STATEMENT
The mice experiments were performed with the approval of the Com-

(
Macherey-Nagel GmbH & Co. KG, Düren, Germany) according to the instructions.RNA quality and quantity were measured with NanoDrop spectrophotometer (Thermo Fischer Scientific).The quality of total RNA was further verified with Agilent 2200 TapeStation and RNA ScreenTapes (Agilent Technologies, Santa Clara, CA).RNA sequencing (RNA-Seq) service was performed by the Biomedicum Functional Genomics Unit (HiLIFE and Biocenter Finland, University of Helsinki, Helsinki, Finland) using mRNA purification with polyA-binding beads, NEBNext Ultra II Directional RNA Library Prep kit (New England Bio-Labs, Ipswich, MA) and single-end sequencing (read length 75 bp) using Illumina NextSeq sequencer (Illumina, San Diego, CA) with a High Output run.Four biological replicates were analyzed.The data quality was first analyzed with FastQ and summarized with MultiQC, light quality trimming was done with Trimmomatic and the sample reads were aligned against the human GRCh38.p12reference genome with STAR, with mapping quality assessment done with Qualimap.To quantify the aligned reads, featureCounts was used.Finally, differential expression statistics were calculated with DESeq2 and basic gene annotation was obtained using Ensembl release 97.The sequencing coverage and quality statistics for each sample are summarized in Table

( 1 . 1 -
and 1.2-fold, respectively) in GNASmt cells without any effect on the Par cells.As a positive control, 5 μM staurosporine induced over 3-fold caspase activation in both Par and GNASmt cells (data not shown).

2
Effect of GNAS mutation on expression of PDE4D and PDE4B transcripts in HCT116 cells as analyzed by qPCR.(A) PDE4D expression in Par and GNASmt cells.(B) PDE4B expression in Par and GNASmt cells.(C) PDE4D expression in GNASmt cells after treatment with GNAS siRNAs.(D) PDE4B expression in GNASmt cells after treatment with GNAS siRNAs.Scrambled siRNA was used as a control (Ctrl).The values are means ± SEM from three independent experiments (n = 9).** P < .01,*** P < .001.

7 F
I G U R E 4 Tumor growth of HCT116 cells in vivo, and expression of PDE4D transcript in human colorectal cancer (CRC) specimens.(A) Growth curves of Par and GNASmt tumors in nude mice.The data are presented as mean ± SEM (* P < .05;eight mice in both groups).(B) Expression of PDE4D mRNA (ENSG00000113448) in human CRC samples (n = 35) with or without GNAS mutations.The horizontal line denotes the median among GNAS wild-type tumors; the first and third quartiles are marked with dashed lines.

F
I G U R E 5 cAMP synthesis, expression of PDE4D and PDE4B transcripts, and proliferation of GNAS-mutated SK-CO-1 CRC cell line.(A) cAMP accumulation in Par and GNASmt HCT116 cells and in SK-CO-1 cells as measured with cAMP Glo Max assay.(B) Expression of PDE4D mRNA in Par and GNASmt HCT116 cells and in SK-CO-1 cells as analyzed with qPCR.(C) Expression of PDE4B mRNA in Par and GNASmt HCT116 cells and in SK-CO-1 cells as analyzed with qPCR.(D) Expression of PDE4D mRNA in SK-CO-1 cells after treatment with GNAS siRNAs.(E) Expression of PDE4B mRNA in SK-CO-1 cells after treatment with GNAS siRNAs.Scrambled siRNA was used as a control (Ctrl) in D and E. (F) Proliferation of SK-CO-1 cells with and without PDE4 inhibitors (Ro 20-1724, 10 μM and GEBR-7b, 10 μg/mL) or FSK (10 μM) as analyzed with CellTiter Blue Cell Viability Assay.The values are means ± SEM from three independent experiments, except two in A (n = 12 in A, n = 9 in B-E, n = 18 in F). * P < .05,** P < .01,*** P < .001.
have reported GNAS knockout to decrease in vitro colony formation and organoid growth of the inherently GNAS mutant CRC cell lines KM12, SNU175 and SK-CO-1.Furthermore, they reported that overexpression of either wild-type or mutant GNAS increased in vitro growth of LS174T CRC cells.One potential explanation for these opposite results on CRC cell growth after introduction of GNAS mutation is that More et al used a vector-based system utilizing tetracycline response element promoter, whereas we knocked in a gain-of-function mutation to the endogenous gene using CRISPR/Cas9 technology.These two distinct techniques may lead to different steadystate expression levels of the mutated protein, which could explain subsequent effects on cell growth.Further, More et al analyzed in vitro growth using colony formation assay, whereas we measured cell viability and thymidine incorporation.These assays measure distinct properties of 2D cell growth and survival.In vivo, More et al26 have reported GNAS knockout to decrease the peritoneal growth of the GNASmutated CRC cell line KM12, and overexpression of wild-type or mutant GNAS to increase the growth of LS174T CRC cells.Our in vivo data show that Par HCT116 cells formed larger tumors than the GNASmutated ones in a subcutaneous nude mouse model.In line with our results, PDE4D knockdown has been shown to reduce subcutaneous mittee for Animal Experiments of the District of Southern Finland (ESAVI/22896/2020, ESAVI/10262/2022).The study of human tumors has been reviewed and approved by the Ethics Committee of the Hospital District of Helsinki and Uusimaa (HUS/2509/2016) and the Finnish Institute for Health and Welfare (THL/1300/5.05.00/2019).All samples were collected after informed consent.
Expression of cAMP-hydrolyzing phosphodiesterases in parental and GNAS-mutated HCT116 cells as detected by RNA sequencing.
mutated (GNASmt) cells.(B) cAMP accumulation in Par and GNASmt cells after treatment with GNAS siRNAs.Scrambled siRNA was used as a control (Ctrl).The values are means ± SEM from two independent experiments (n = 8).** P < .01,*** P < .001.T A B L E 1 a Benjamini-Hochberg adjusted.b Cleaving both cAMP and cGMP.